IntegrityLens
All work
1st prizeHackathon · Web · AI · 2024

AI that links officials’ luxury items to their income declarations

IntegrityLens is a web app for investigative journalists. You upload a photo of a public official, image recognition finds the luxury item and its price, and the app sets it against what that official declared as income. Created at the EaP Hackathon 2024 in Moldova, where it won the EaP Civic Tech Award.

Event
EaP Hackathon, Moldova
Year
2024
My role
Product design · UX writing
Platform
Web app
IntegrityLens on desktop. Left: the investigation form with a name, an institution and an uploaded photo. Right: the result, with a warning that one item costs more than the official's monthly salary, the declared salary, outgoings and cash, and three similar bags with prices and similarity scores.

What I did

Hackathon · Concept to prototype

I designed the product experience: the investigation flow, how a result is shown, and the dossiers where findings are kept. I also wrote the interface copy, so that a sensitive finding reads as evidence to check, not as an accusation.

  • Investigation flow
  • Result + risk flag
  • Dossiers
  • UX writing
The problem

The photo is public. So is the salary.

Public officials in Georgia publish asset declarations on declaration.acb.gov.ge. They also appear in public and in photos, sometimes wearing things that look expensive. Both facts are open, but nobody connects them by default.

For a journalist, connecting them is slow manual work: name the item, find what it costs, open the declaration, do the maths. IntegrityLens does that part in one step, so the journalist can spend their time on the actual investigation.

How it works
  1. Name the official and their institution, then upload one photo.
  2. Image recognition finds similar products with their prices and a similarity score.
  3. The app compares them with the declaration and flags any item that costs more than a month’s salary or 10% of declared cash.
The flag in the example

$4,500

One bag, against a declared salary of 5,500 GEL a month. The app shows the gap. The journalist decides what it means.

Investigate

From one photo to a checkable finding

Three fields on the left, the answer on the right. The declared numbers sit above the matches, so the price of each item is always read against the official’s own income.

Investigation form: Who do you want to investigate? Fields for full name, institution and a photo upload, with a note that results may not be accurate and additional research may be required.
01 · Who and what to check
Result panel: a yellow warning about an item costing more than the monthly salary or 10% of cash, the declared salary, outgoings and cash with a link to the full declaration, and three similar bags with prices, similarity scores and checkboxes, then Download summary and Add to dossier.
02 · Matches against the declaration
Design decisions

Careful with a sensitive claim

Dossiers

One place per official, built up over time

Investigations rarely end with one photo. A dossier keeps every item found for one official, the original photo, private notes and the matches, with a summary to download and share with an editor.

Dossiers tab: a list of officials with their role and institution, the selected one showing declared salary, outgoings and cash. On the right, chips for each item found, the photo with a notes field, similar items with prices and a Download summary button.
Outcome

1st prize at the EaP Civic Tech Award 2024

IntegrityLens won first place at the EaP Hackathon in Moldova, recognised for supporting anti-corruption work by journalists and civil society. See the LinkedIn post ↗